Calcium Waves Precede Electrophysiological Changes of Spreading Depression in Hippocampal Organ Cultures

Electrophysiological behavior of HOTC

The general histological organization of HOTCs (Fig.1A) is consistent with previous reports of HOTCs (Zimmer and Gähwiler, 1984;Frotscher and Gähwiler, 1988). The HOTC shown is from a culture maintained in vitro for 25 d. HOTCs used in this study were maintained in vitro for 21–35 d to help ensure that synaptic activity had reached a plateau before experimental manipulations (Buchs et al., 1993; Muller et al., 1993; Bahr, 1995). All cultures showed distinguishable CA1, CA3, and DG subfields. The CA3 pyramidal cell layer maintained a compact layer similar to that seenin vivo. In contrast to that seen in vivo, the CA1 pyramidal cells spread out to form a loosely defined layer (Kunkler and Kraig, 1997a). In addition, the infrapyramidal limb of the DG frequently was diminished in width (Zimmer and Gähwiler, 1984;Kunkler and Kraig, 1997a), yet overall the cytoarchitectural organization of HOTCs was comparable to that seen in vivo.

Evoked synaptic behavior of the CA3 subfield was also similar to that found in vivo and in situ (Fig.1B). Hippocampal mossy fibers arise from DG granule cells and terminate on CA3 pyramidal cells in vivo (Lorente de Nó, 1934; Hamlyn, 1962; Blackstad et al., 1970; Ramon y Cajál, 1995). Similar synaptic connections from the DG to the CA3 pyramidal cells exist within HOTCs (Zimmer and Gähwiler, 1984). Accordingly, bipolar electrical stimulation of the DG may be expected to show evoked field potentials within HOTCs that are similar to those seen in vivo. Field potential recordings (Fig.1B) of CA3 pyramidal cell responses to DG stimulation confirmed this. With sufficient stimulation of the mossy fiber afferents, CA3 pyramidal cells began to fire action potentials noted in the interstitial space as a population spike (a, Fig.1B). This was seen during the longer-lasting field EPSP (fEPSP) (b, Fig. 1B) that was maximally inverted (i.e., moves in the positive direction) at the pyramidal cell layer (trace 4, Fig. 1B), where synaptic currents from excitatory dendritic synapses flow from pyramidal cells nearer the cell body (Andersen et al., 1971a,b; Martin, 1983; Alger et al., 1984). Stimulus strength was increased so that the mossy fiber presynaptic volley that otherwise would have been evident (i.e., at a, Fig. 1B) was, instead, overshadowed by the pyramidal cell population spike (most evident intraces 3 and 4, Fig. 1B). Field potentials commonly were recorded for SD experiments at the CA3 pyramidal cell layer. To prevent possible induction of synaptic facilitation or depression (Malenka, 1994), we evoked field potentials at a rate of 0.5 Hz for no more than 2–3 min every 10–15 min. In most experiments a single field potential was evoked to verify microelectrode placement and the physiological status of the HOTC before the induction of each SD.

Electrographic seizures could be induced in HOTCs (Fig.2). Spontaneous fast electrical activity often was seen when DC potential microelectrodes were advanced into the interstitial space of HOTCs. Such spontaneous electrical activity could be enhanced to include electrographic seizures by, for example, a single bipolar electrical stimulus at the hilus. Others have found similar electrophysiological behavior within HOTCs (Fowler et al., 1986). Ideally, SD can be studied best when evoked by predetermined stimuli, such as bipolar stimulation in the hilus. However, SD will not pass through gray matter areas that are hyperexcitable (i.e., are experiencing or recently have experienced electrographic seizures) (Bureś et al., 1975; Koroleva and Bureś, 1979, 1980). Electrographic seizures within HOTCs also prevented SD.

The occurrence of electrographic seizures usually was prevented by the addition of ISO, a β-adrenergic agonist, along with raising Ca²⁺ from 1.25 to 2.5 mm in all Ringer’s solutions. This effect of ISO might seem paradoxical, because ISO increases pyramidal cell excitability. For example, ISO increases population spike amplitude in adult CA1 pyramidal cells (Mueller et al., 1981; Dunwiddie et al., 1992) without affecting the fEPSP (Heginbotham and Dunwiddie, 1991). ISO affected HOTCs in this manner. In addition, ISO depolarizes and increases the input resistance of CA3-affected pyramidal cells in immature hippocampal slices as young as postnatal day 7 (Moudy and Schwartzkroin, 1992). However, ISO also increases gap junctional connectivity among astrocytes (Giaume et al., 1991). Thus, the ability of ISO to enhance pyramidal cell excitability and yet decrease electrographic seizures may be attributable, at least in part, to direct effects of this β-adrenergic agonist onastrocytic gap junctional connections, which could increase the dispersal of K⁺ via increased spatial buffering (Newman, 1995). Nonetheless, despite the improved electrophysiological function after the addition of ISO to the normal Ringer’s solution that was bathing HOTCs, SD still could not be induced by bipolar recurrent electrical stimulation in the hilus.

Induction of SD in HOTCs

SD could be induced by hilar bipolar electrical stimulation when Cl⁻ in the Ringer’s solution was replaced by other anions (Bureś et al., 1974). Cl⁻ was reduced by (1) removing NaCl (124 mm) from the Ringer’s solution, (2) replacing 124 mm NaCl with an isosmotic amount of sucrose, (3) replacing 124 mm NaCl with an equimolar amount of sodium propionate or (4) sodium acetate (NaAc) (Bureś et al., 1974). At least three different HOTCs were used for each Ringer’s solution above, and the cultures were exposed/stimulated at least 10 times. Each of these Ringer’s solutions provided a sufficient “conditioning” change to induce SD after a single bipolar electrical stimulus in the hilus. However, substitution with NaAc was the most consistent conditioning change. Therefore, it was used to induce SD in all experiments. Prolonged exposure of HOTCs to NaAc–Ringer’s alone would induce electrographic seizures. However, with the addition of ISO and raised Ca²⁺ to all Ringer’s solutions, coupled to transient (90–120 sec) exposure of HOTCs to NaAc–Ringer’s solution, seizure activity was rarely seen. Finally, SD was induced consistently (n > 100) only when Ringer’s solution above the HOTCs was removed and replaced with mineral oil or moist air. Only rarely would exposure to NaAc–Ringer’s successfully induce SD in submerged HOTCs. None of the other anion-substituted Ringer’s solutions above were a sufficient conditioning solution for SD when HOTCs were submerged.

Two of the three fundamental electrophysiological characteristics of SD are shown in Figure 3. The first is the slow interstitial DC potential change of SD (thick record, Fig. 3). Exposure to NaAc–Ringer’s always produced a small positive deflection in the interstitial DC potential. After 90–120 sec the Ringer’s solution was returned to normal, and a single bipolar electrical stimulus was delivered quickly to the hilus. This invariably produced SD, as evidenced by a dramatic negative shift in the interstitial DC potential of 30–50 mV (points 2–4 of slow DC potential record, Fig. 3) (Leão, 1944). When the DC shift of SD began to recover toward the baseline, spontaneous epileptiform activity commonly was seen (points 5–7 along DC potential record, Fig. 3). This has been observed in hippocampal brain slices derived from adult (Snow et al., 1983) and immature (Psarropoulou and Avoli, 1993) rats. Notice that the bipolar stimulus to trigger SD also is associated with a uniquely large DC negative shift (between points1 and 2 of the slow DC potential record, Fig. 3). This initial negative deflection of ∼10 mV occurred synchronously throughout the HOTC, as evidenced by three microelectrodes separated by 100 μm (Fig. 4) and placed variably throughout HOTCs. Furthermore, the initial DC negative deflection was never evident from bipolar stimulation without exposure to NaAc–Ringer’s (data not shown). The DC potential change from the bipolar stimulus also was associated with a fast propagating Ca²⁺ wave (see below). These fast Ca²⁺ waves were never seen with bipolar stimuli in HOTCs exposed to normal Ringer’s solution.

The second hallmark of SD, transient cessation of evoked electrical activity (thin traces 2–4, Fig. 3) (Leão, 1944), was also evident during the negative, slow DC potential change. CA3 pyramidal cell responses could become hyperactive after successive SDs (n = 12). This was evident before SD (fast DC potentialtrace 0, Fig. 3) in which two population spikes are seen before exposure to NaAc–Ringer’s. The latter further increased excitability (fast DC potential trace 1, Fig. 3) so that multiple population spikes occurred along with a dramatically widened fEPSP. Then, as expected with the onset of SD (fast DC potentialtrace 2, Fig. 3) until the peak slow DC change (fast DC potential traces 3 and 4, Fig. 3), no evoked field potentials were seen. With recovery (fast DC potentialtraces 5–8, Fig. 3), pyramidal evoked activity progressively returned. Initially, only an exaggerated fEPSP could be seen (fast DC potential traces 5 and 6, Fig. 3). This quickly was followed by epileptiform activity (as noted above in the slow DC potential record) and the production of multiple population spikes by bipolar stimulation (fast DC potential trace 7, Fig. 3). Finally, the fEPSP was reduced immediately after SD (compare fast DC potential traces 0 and 8, Fig. 3).

The third classical electrophysiological change of SD, slow propagation of a large negative interstitial DC potential (Leão, 1944), is shown in Figure 4. DC propagation velocities averaged 5.47 ± 0.47 mm/min (n = 15), a speed similar to values foundin vivo (Leão, 1944; Bureś et al., 1974) and in the in vitro hippocampal brain slice preparation (Snow et al., 1983; Psarropoulou and Avoli, 1993).

Intercellular Ca²⁺ waves and SD

Glutamate can be an effective stimulus to evoke SD (Bureś et al., 1974). Furthermore, it evokes propagating intracellular and intercellular Ca²⁺ waves in HOTCs (Dani et al., 1992) (Fig. 5). HOTCs are surrounded by a rim of glial (astrocytes and microglia) cells (Kunkler and Kraig, 1997a) that can be used as a mixed glial cell preparation. Ca²⁺ waves from glutamate exposure were noted in both of these areas. In the mixed glial cell rim around the HOTCs, Ca²⁺ waves could be seen in individual cells (Fig.5). Glutamate (100 μm) evoked oscillations within single cells as well as rapidly propagating waves of Ca²⁺between cells (n = 10), as has been reported previously in primary astrocytic cultures (Cornell-Bell et al., 1990; Charles et al., 1991, 1992; Cornell-Bell and Finkbeiner, 1991; Finkbeiner, 1992;Enkvist and McCarthy, 1994). Ca²⁺ waves also were seen within HOTCs during exposure to elevated glutamate. These changes have been noted previously within HOTCs (Dani et al., 1992) and within the in situ rat retina (Newman and Zahs, 1997). However, electrophysiological changes of SD were never seen from glutamate exposure (n = 30). Thus, glutamate-evoked Ca²⁺ waves can occur without electrophysiological evidence of SD.

Electrophysiological changes of SD, on the other hand, do occur with Ca²⁺ waves (Fig. 6). Transient exposure to NaAc–Ringer’s was necessary to elicit SD in HOTCs by a single bipolar stimulus to the hilar area of the DG. SD always (n = 6) occurred with two distinct Ca²⁺ waves. The first occurred within 1–2 sec of the synchronous DC deflection recorded in the pyramidal cell layer from bipolar electrical stimulation. Furthermore, it propagated >100 μm/sec in the oriens layer and originated either in CA1 or CA3. Figure 6 shows this wave beginning in CA1/CA2 (image 1, Fig.6) and rapidly propagating toward the hilus along the oriens layer (image 2, Fig. 6). The second wave propagated mostly perpendicular to the pyramidal cell layer from a nidus in CA3 (or less often CA1) (images 3–6, Fig. 6). In addition, this second wave propagated at the speed of SD (4.0 ± 0.2 mm/min;n = 6). However, the peak change in Ca²⁺ of this second wave always preceded the peak interstitial DC change of SD by a significant (p< 0.001 tested by an unpaired Student’s t test;n = 6) margin of 6–16 sec.

To explore further whether Ca²⁺ waves are essential for SD initiation and propagation, we exposed HOTCs to heptanol (data not shown). Heptanol reversibly uncouples gap junctions (Spray and Bennett, 1985; Bastiaanse et al., 1993). Furthermore, it can stop the propagation of intercellular Ca²⁺ waves as well as the initiation of SD (Martins-Ferreira and Ribeiro, 1995; Nedergaard et al., 1995). DC potentials were recorded in the CA1 and CA3 pyramidal cell layers after HOTCs were loaded with Fluo-3. Ethanol, the vehicle for heptanol, did not affect evoked field potentials or SD and associated Ca²⁺ waves (n = 5). Heptanol in high concentrations (3 mm) reversibly abolished field potentials (n = 3) as well as Ca²⁺ waves and SD. However, because an evoked field potential was needed to initiate SD, the dose of heptanol was titrated downward to discover if Ca²⁺ waves could be separated from SD. The results suggest that such a separation cannot be accomplished. First, without exposure to heptanol, electrophysiological changes of SD occasionally did not propagate from CA3 to CA1 (and vice versa). Under these cases, Ca²⁺ waves were not seen in those areas that did not show DC changes of SD (n = 7). Then in other experiments, exposure to lower levels of heptanol (0–1.5 mm) often extinguished Ca²⁺waves and SD despite preserved evoked field potentials in both CA3 and CA1 (n = 9 HOTCs and 26 SDs). For example, SD and Ca²⁺ waves were seen in 11 of 13 trials (85%) at 0.75 mm heptanol, 2 of 4 trials (50%) at 1.0 mm heptanol, 2 of 9 trials (22%) at 1.5 mmheptanol, and 0 of 3 trials at 3.0 mm heptanol. Thus, two distinct types of Ca²⁺ waves precede SD and appear to be essential for the initiation and propagation of this phenomenon.

Initiation of SD in HOTCs

Despite the increased use of HOTCs, SD has not been reported in these cultures. This lack is attributable to at least two facts. First, although acute rodent hippocampal slices easily support SD (Snow et al., 1983; Psarropoulou and Avoli, 1993), HOTCs typically do not. Both preparations lack extrinsic inputs. Furthermore, they both display similar spontaneous and evoked electrophysiological activity. Nonetheless, the tissue volume activated by depolarizing stimuli is less in HOTCs than in acute hippocampal brain slices. HOTCs are ∼50–150 μm thick (our unpublished observations), whereas acute hippocampal slices are typically 300–400 μm thick (Alger et al., 1984). Thus, DG gyrus bipolar electrical stimulation activates a smaller volume of CA3 pyramidal cells in HOTCs than acute slices. SD initiation requires a sufficient volume of gray matter to be acutely depolarized (Bureś et al., 1974). Perhaps the volume of CA3 activated by bipolar DG stimulation in HOTCs is simply too low to initiate SD in either CA3 or CA1 under normal conditions. This conclusion is supported by our finding that SD could be initiated very rarely in HOTCs submerged under normal Ringer’s superfusion. However, when HOTCs were used in an “interface” chamber where Ringer’s exposure was restricted to the bottom surface of the cultures, SD could begin to be initiated by electrical stimulation. This implies that soluble substances released to the interstitial space are essential for SD. Diffusion (Nicholson, 1995) of essential materials away from thin submerged HOTCs might be too rapid to allow for SD initiation. Alternatively, ephaptic interactions between neural cells (Taylor and Dudek, 1984; Faber and Korn, 1989) could be enhanced within HOTCs in an interface chamber, as compared with submerged HOTCs.

A second potential reason why SD has not been reported previously in HOTCs stems from the seemingly convoluted “conditioning” needed to evoke the phenomenon consistently. Our first attempts at inducing SD in HOTCs centered on increasing HOTC excitability by altering Ringer’s constituents (our unpublished observations) and were unsuccessful. Raising Ringer’s K⁺ or lowering Ca²⁺ or Mg²⁺ only induced spontaneous or evoked seizure activity. Similarly, the addition of glutamate, NMDA, carbachol, or 4-aminopyridine, each of which has excitatory effects, only resulted in seizure activity. Later attempts centered on reducing Ringer’s Cl⁻. This is a well recognized method of altering interstitial fluid to promote the induction of SD (Bureś et al., 1974). The mechanisms responsible for this effect are unknown but may be related to increased excitability (Nicholson and Kraig, 1981; Haglund and Schwartzkroin, 1984). Again, only epileptiform activity occurred when Cl⁻ was replaced with propionate or acetate. Similar results occurred in acute hippocampal slices when Ringer’s Cl⁻ was replaced with propionate (Yamamoto and Kawai, 1967). The final, and virtually always successful, method used to induce SD involved transient exposure to NaAc–Ringer’s that also contained ISO and raised Ca²⁺.

Ca²⁺ waves and SD

Gap junctions between neurons (Somjen et al., 1992; Herreras et al., 1994; Largo et al., 1997) and astrocytes (Nedergaard et al., 1995) may play important roles in SD initiation and propagation, respectively. Although raised Ca²⁺ has no direct effect on gap junction, other Ringer’s changes (the addition of ISO and the replacement of Cl⁻ by acetate) could increase gap junctional conductances. For example, ISO increases gap junctions between cultured astrocytes (Giaume et al., 1991). Furthermore, transient exposure to NaAc should begin to acidify neural cells (Roos and Boron, 1981). This might be expected to reduce gap junctional conductances (Spray et al., 1981). However, conductance through gap junctions made of connexin 43 and connexin 32, gap junctional proteins found in brain (Giaume and McCarthy, 1996), increases with a reduction in cellular pH. For example, gap junctions formed between pairs of Xenopus laevis oocytes expressing homologous channels for these proteins show an increase in conductance when intracellular pH is lowered from 7.2 to 6.8 or from 7.2 to 6.3, respectively (Liu et al., 1993).

The magnitude of the interstitial DC potential change from bipolar stimulation (see Fig. 3) supports the notion of increased cellular connectivity during HOTC exposure to NaAc. During normal Ringer’s solution exposure the DC deflection from a bipolar stimulus to the DG was not seen in CA3 (i.e., points 0 and 8 of slow DC potential record, Fig. 3) or CA1 (data not shown). However a positive and then a large negative DC deflection of ∼10 mV was always seen in CA3 (and CA1) synchronously with the same stimulus during NaAc exposure. The large size of this DC deflection and the distance (i.e., >1 mm) over which it simultaneously was seen implies the existence of a long-distance conductance path along the pyramidal cell layer. This could be attributable to the opening of normally closed neuronal gap junctions between pyramidal cells, as has been suggested for seizures (Perez-Velazquez et al., 1994) and SD (Somjen et al., 1992; Herreras et al., 1994; Largo et al., 1997).

The speed and cytoarchitectural direction of the rapidly propagating Ca²⁺ wave seen shortly after the start of SD further support the suggestion of increased gap junctional connectivity among pyramidal cells. Neuronal Ca²⁺ waves travel at >100 μm/sec (Charles et al., 1996), a speed consistent with that seen for the rapidly propagating wave that spreads along the pyramidal cell basilar dendritic layer (see Fig. 6). Furthermore, no cellular elements extend along this layer that otherwise could account for the propagating intercellular Ca²⁺ wave. Finally heptanol, which uncouples gap junctions (Spray and Bennett, 1985;Bastiaanse et al., 1993), stopped this Ca²⁺ wave as well as the large DC deflection associated with bipolar stimulation.

Heptanol also stopped the slower Ca²⁺ wave and SD itself, as reported by others (Martins-Ferreira and Ribeiro, 1995;Nedergaard et al., 1995). The slowly propagating Ca²⁺ wave traveled at speeds consistent with similar waves seen among astrocytes (Cornell-Bell et al., 1990; Charles et al., 1991, 1992; Cornell-Bell and Finkbeiner, 1991; Finkbeiner, 1992;Enkvist and McCarthy, 1994). Furthermore, the slower Ca²⁺ wave spread mostly perpendicular to pyramidal cells but also in all directions from an initial focus. Astrocytes are dispersed ubiquitously throughout the laminae of the HOTC (Kunkler and Kraig, 1997a). Therefore, as suggested by others (Nedergaard, 1994;Nedergaard et al., 1995), it is plausible to propose that this slower Ca²⁺ wave during SD in HOTCs also relies on gap junctional connections among astrocytes. A Ca²⁺ wave that precedes SD has been noted by others in acute hippocampal brain slices (Basarsky and MacVicar, 1997).

Significance of SD in HOTCs

The discovery of two distinct Ca²⁺ waves that precede electrophysiological changes of SD adds further support to the suggestion that changes in neural function precede the regenerative depolarization of SD (Herreras et al., 1994). Somjen and colleagues (Somjen et al., 1992; Herreras et al., 1994; Largo et al., 1997) suggest that SD involves the opening of previously closed neuronal gap junctions. That these waves occur before the regenerative electrophysiological changes of SD suggest they could effect triggering mechanisms within (Post and Silberstein, 1994; Finkbeiner, 1995) and among neural cells (Nedergaard, 1994; Parpura et al., 1994) that are essential for SD. Furthermore, given the diverse and potentially long-lasting effects of cellular Ca²⁺ changes (Whitfield, 1990; Post and Silberstein, 1994), Ca²⁺waves from SD also may trigger mechanisms responsible for temporally remote phenomena, such as SD-induced modulation of ischemic injury.

The discovery of the mechanisms responsible for SD has increased importance because SD is now known to influence ischemic brain injury. When SD precedes ischemia by <1 d, injury is greater than otherwise would be expected from ischemia alone (Takano et al., 1996). However, when SD occurs >1 d before ischemia, injury is reduced from that which otherwise would be seen (Kawahara et al., 1995; Kobayashi et al., 1995;Matsushima et al., 1996). Clarification of the mechanisms responsible for SD could lead to improved insight to the pathogenesis of ischemic brain injury. Furthermore, knowledge of specific mechanisms responsible for SD might reveal how brain injury from ischemia may be mitigated or perhaps even reversed. Currently, the mechanisms by which SD confers this dual effect on ischemic injury are unknown. However, HOTCs may be ideal for studies directed toward such mechanisms, because HOTCs support SD and they can be maintained in vitro for months, where individual cells can be followed in space and time.